GPT-4
OpenAI's GPT 4: a large multimodal model that accepts image and text inputs. Features predictable scaling, OpenAI Evals, and advanced reasoning.
Model Overview
Capabilities, design details, and architectural traits
GPT 4 - A Large Multimodal Model With Image and Text Input
GPT 4 is described by OpenAI as a large multimodal model that accepts image and text input and produces text output only. It marks a specific point in OpenAI's own development process, where training behavior became something the company could predict ahead of time.
| Trait | Detail |
|---|---|
| Input and output split | Accepts both image and text as input, but generates text output only, unlike later models built for output across multiple modalities. |
| Predictable training run | Documented by OpenAI as its first large model whose training performance could be accurately predicted ahead of time, based on lessons from an earlier GPT-3.5 test run. |
| Six month alignment period | OpenAI states it spent six months iteratively aligning GPT 4 using lessons from adversarial testing and ChatGPT, aimed at factuality, steerability, and guardrail adherence. |
| Staged image rollout | At launch, text input shipped broadly through ChatGPT and the API with a waitlist, while image input was prepared for wider availability through a single partner first. |
Released Alongside OpenAI Evals
OpenAI open sourced OpenAI Evals alongside GPT 4, a framework for automated evaluation of AI model performance, built so anyone could report shortcomings and help guide further improvements.
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | OpenAI | Anthropic | Anthropic |
| Release Date | March 14, 2023 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | Sep 2021 | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 8K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $30 | $4 | $2 Best Input Pricing |
| Output Pricing | $60 | $20 | $10 Best Output Pricing |
| Modalities | |||
| Inputs | text | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 6.7 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | 13.1 | - | - |
| Agentic Index | - | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from OpenAI
Other models by OpenAI
Top AI Models
Leading alternatives by intelligence score
